🚗 CrashSenseAI

Senses Crashes Instantly with AI

---

🔥 Overview

CrashSenseAI is a real-time accident detection system using Edge AI models.
It detects crashes from videos or streams, raises sound alarms, saves accident frames, and instantly sends email notifications.

🛠 Built With

• Python 3.10.0
• YOLOv8 (Ultralytics) — Deep Learning Object Detection
• OpenCV — Video processing
• PyTorch — Neural Network backend
• Tkinter + ttkbootstrap — Clean Dark GUI
• Pygame — Sound alarms
• Automated Email — With crash frame attachments

---

🚀 Features

• 🎯 Real-time accident detection
• 💾 Auto-save accident frames in organized folders
• 📩 Auto-send email alerts with crash images
• 🔊 Sound alarm on accident detection (Mute/Unmute option)
• 🎥 Original video playback speed (no artificial slowdown)
• 🖥️ Clean and modern dark GUI (Responsive and Lightweight)
• ⚡ Easy setup, lightweight, and customizable

---

📦 Setup Instructions

1. Clone the Repository

git clone https://github.com/111221007/CrashSenseAI.git
cd CrashSenseAI

2. Install Dependencies

pip install -r requirements.txt

requirements.txt includes:

• opencv-python
• numpy
• torch
• ttkbootstrap
• pygame
• Pillow
• ultralytics

---

3. Prepare Required Folders and Files

✅ Place your trained YOLO model at:

CrashSenseAI/accident_detection/model/best.pt

✅ Add your test videos inside:

CrashSenseAI/accident_detection/data/input/test_videos/

✅ Add an alarm sound file (e.g., alarm.mp3) here:

CrashSenseAI/accident_detection/data/alarm.mp3

✅ Setup your email credentials inside email_sender.py:

sender_email = "your_email@gmail.com"
sender_password = "your_app_password"
receiver_email = "receiver_email@gmail.com"

⚡ If using Gmail, you must enable App Passwords to use email sending. Learn how to create App Password →

---

▶️ How to Run CrashSenseAI

1. Launch the GUI:

   python gui.py

2. Select a test video from the GUI.

3. Click Start Detection.

The system will:

• Detect crashes in real-time
• Save crash frames inside:

  accident_detection/data/output/accident_frames/SelectedVideo/
  

• Play sound alarms when crash detected
• Send automatic emails with saved crash images attached
• Allow you to Mute/Unmute alarms anytime from the GUI

---

📂 Project Structure

CrashSenseAI/
├── accident_detection/
│   ├── data/
│   │   ├── input/test_videos/       # ➡️ Your test videos
│   │   ├── output/accident_frames/  # ➡️ Saved crash frames
│   │   └── alarm.mp3                # ➡️ Alarm sound
│   ├── model/
│   │   └── best.pt                  # ➡️ YOLOv8 trained model
│   ├── scripts/
│   │   └── gui.py                   # ➡️ GUI Application
│   └── email_sender.py              # ➡️ Email automation
├── requirements.txt
└── README.md

---

🌟 Future Roadmap

• 🔴 Live webcam & YouTube live stream crash detection
• 📊 Real-time dashboard monitoring accidents
• ☁️ Cloud-based remote alert system (IoT Integration)
• 📈 Performance optimization for embedded devices (Raspberry Pi / Jetson)

---

👨‍💻 Developed By

Chandramohan Reddy Poreddy
👉 Feel free to contribute by creating a pull request!

---

📜 License

This project is licensed under the MIT License.
Feel free to use, modify, and share responsibly.

---

⭐ If you find this project helpful, please consider giving a star on GitHub! ⭐